GPT-6 Astra
gpt-astra gpt-6-astra $10.00 in / 1M tokens · list
$50.00 out / 1M tokens · list
// server models
Turn your skill into an app Server Models
The tiers a hosted app can run on SkillSafe's servers, billed per token in credits — $1 = 10,000 credits — at the provider's list price, passed through with no markup; SkillSafe charges separately for the computation. The tiers come first; below them, every model grouped by what it takes in and gives out. The models that run in the browser are free.
Latest model of each tier
Provider list price · USD per 1M tokensGPT-6.1 Sol
gpt-sol gpt-6.1-sol $2.00 in / 1M tokens · list
$10.00 out / 1M tokens · list
GPT-5.6 Terra
gpt-terra gpt-5.6-terra $2.00 in / 1M tokens · list
$12.00 out / 1M tokens · list
GPT-6 Luna
gpt-luna gpt-6-luna $0.10 in / 1M tokens · list
$0.50 out / 1M tokens · list
Gemma 4 26B (Workers AI)
gemma-fast @cf/google/gemma-4-26b-a4b-it $0.10 in / 1M tokens · list
$0.30 out / 1M tokens · list
GPT Image 2.5 Flare
gpt-image gpt-image-2.5-flare ~$0.060 list, per 1024×1024 image
FLUX.2 Klein 9B (Workers AI)
flux-klein @cf/black-forest-labs/flux-2-klein-9b ~$0.015 list, per 1024×1024 image
Deepgram Aura 2 English (Workers AI)
aura @cf/deepgram/aura-2-en $0.030 list, per 1k characters
Whisper Large v3 Turbo (Workers AI)
whisper @cf/openai/whisper-large-v3-turbo $0.000513 list, per audio minute
Model cost is pass-through: these are the providers' published list prices, unchanged — see OpenAI pricing and Workers AI pricing. What SkillSafe charges for is the computation and API handling on its own servers: a 10% platform fee on each run's base — the model's list cost, with no per-job overhead. A creator's markup is a separate cut of the same base — they keep 100% of it. Only the latest model of each tier is shown; configure the alias and your app follows upgrades. Tiers that cannot run on this deployment are not shown.
Every model, by what goes in and what comes out
52 models · pinned idsText Text 22 models
Chat completion: the app's prompt plus the run input go in, prose comes back.
| Model | Provider | in / out per 1M tokens · list | Caps |
|---|---|---|---|
Granite 4.0 H Micro (Workers AI) @cf/ibm-granite/granite-4.0-h-micro | Workers AI | $0.017 / $0.11 | 100k in · 4k out · 120s |
Llama 3.2 1B (Workers AI) @cf/meta/llama-3.2-1b-instruct | Workers AI | $0.027 / $0.20 | 55k in · 4k out · 120s |
Llama 3.1 8B fp8-fast (Workers AI, legacy) @cf/meta/llama-3.1-8b-instruct-fp8-fast | Workers AI | $0.045 / $0.38 | 27k in · 4k out · 120s |
Llama 3.2 3B (Workers AI) @cf/meta/llama-3.2-3b-instruct | Workers AI | $0.051 / $0.34 | 75k in · 4k out · 120s |
Qwen3 30B-A3B (Workers AI, thinks) @cf/qwen/qwen3-30b-a3b-fp8 | Workers AI | $0.051 / $0.34 | 28k in · 4k out · 120s |
GLM-4.7 Flash (Workers AI) @cf/zai-org/glm-4.7-flash | Workers AI | $0.061 / $0.40 | 100k in · 4k out · 120s |
Mistral 7B v0.1 (Workers AI, legacy, 2.8k context) @cf/mistral/mistral-7b-instruct-v0.1 | Workers AI | $0.11 / $0.19 | 2k in · 1k out · 120s |
Llama 3.1 8B fp8 (Workers AI) @cf/meta/llama-3.1-8b-instruct-fp8 | Workers AI | $0.15 / $0.29 | 27k in · 4k out · 120s |
GPT-OSS 20B (Workers AI, reasoning) @cf/openai/gpt-oss-20b | Workers AI | $0.20 / $0.30 | 100k in · 4k out · 120s |
Llama 3.1 8B (Workers AI, legacy, 8k context) @cf/meta/llama-3.1-8b-instruct | Workers AI | $0.28 / $0.83 | 3k in · 4k out · 120s |
Llama 3.3 70B fp8-fast (Workers AI) @cf/meta/llama-3.3-70b-instruct-fp8-fast | Workers AI | $0.29 / $2.25 | 19k in · 4k out · 120s |
Llama 3.1 70B fp8-fast (Workers AI, legacy) @cf/meta/llama-3.1-70b-instruct-fp8-fast | Workers AI | $0.29 / $2.25 | 19k in · 4k out · 120s |
GPT-OSS 120B (Workers AI, reasoning) @cf/openai/gpt-oss-120b | Workers AI | $0.35 / $0.75 | 100k in · 4k out · 120s |
Gemma SEA-LION v4 27B (Workers AI) @cf/aisingapore/gemma-sea-lion-v4-27b-it | Workers AI | $0.35 / $0.56 | 100k in · 4k out · 120s |
DeepSeek V4 Flash (Workers AI) @cf/deepseek-ai/deepseek-v4-flash-0731 | Workers AI | $0.44 / $1.32 | 100k in · 4k out · 120s |
DeepSeek R1 Distill Qwen 32B (Workers AI, reasoning) @cf/deepseek-ai/deepseek-r1-distill-qwen-32b | Workers AI | $0.50 / $4.88 | 75k in · 4k out · 120s |
Nemotron 3 Super 120B (Workers AI) @cf/nvidia/nemotron-3-120b-a12b | Workers AI | $0.50 / $1.50 | 100k in · 4k out · 120s |
Qwen2.5 Coder 32B (Workers AI) @cf/qwen/qwen2.5-coder-32b-instruct | Workers AI | $0.66 / $1.00 | 28k in · 4k out · 120s |
QwQ 32B (Workers AI, reasoning) @cf/qwen/qwq-32b | Workers AI | $0.66 / $1.00 | 19k in · 4k out · 120s |
DeepSeek V4 Pro (Workers AI) @cf/deepseek-ai/deepseek-v4-pro-0813 | Workers AI | $1.32 / $3.96 | 100k in · 4k out · 120s |
GLM-5.2 (Workers AI) @cf/zai-org/glm-5.2 | Workers AI | $1.40 / $4.40 | 100k in · 4k out · 120s |
GLM-5.3 (Workers AI, thinks) @cf/zai-org/glm-5.3 | Workers AI | $1.40 / $4.40 | 100k in · 4k out · 120s |
Text + image Text 17 models
The same, and the model can also see images attached to the run with $files. Only models that named the colour of a test image are listed here.
| Model | Provider | in / out per 1M tokens · list | Caps |
|---|---|---|---|
GPT-6 Luna gpt-6-luna gpt-luna | OpenAI | $0.10 / $0.50 | 100k in · 4k out · 60s |
Gemma 4 26B (Workers AI) @cf/google/gemma-4-26b-a4b-it gemma-fast default | Workers AI | $0.10 / $0.30 | 32k in · 4k out · 120s |
GLM-5.3 Flash (Workers AI, thinks) @cf/zai-org/glm-5.3-flash | Workers AI | $0.15 / $0.50 | 100k in · 4k out · 120s |
GPT-5.6 Luna gpt-5.6-luna | OpenAI | $0.20 / $1.20 | 100k in · 4k out · 60s |
GPT-5 mini (retiring 2026-12-11) gpt-5-mini | OpenAI | $0.25 / $2.00 | 100k in · 4k out · 60s |
Llama 4 Scout 17B (Workers AI) @cf/meta/llama-4-scout-17b-16e-instruct | Workers AI | $0.27 / $0.85 | 100k in · 4k out · 120s |
Mistral Small 3.1 24B (Workers AI) @cf/mistralai/mistral-small-3.1-24b-instruct | Workers AI | $0.35 / $0.56 | 100k in · 4k out · 120s |
Qwen 3.8 27B (Workers AI) @cf/qwen/qwen3.8-27b | Workers AI | $0.45 / $3.20 | 100k in · 4k out · 120s |
Kimi K2.5 (Workers AI, legacy) @cf/moonshotai/kimi-k2.5 | Workers AI | $0.60 / $3.00 | 100k in · 4k out · 120s |
Kimi K2.6 (Workers AI) @cf/moonshotai/kimi-k2.6 | Workers AI | $0.95 / $4.00 | 100k in · 4k out · 120s |
Kimi K2.7 Code (Workers AI, thinks) @cf/moonshotai/kimi-k2.7-code | Workers AI | $0.95 / $4.00 | 100k in · 4k out · 120s |
GPT-5.1 gpt-5.1 | OpenAI | $1.25 / $10.00 | 100k in · 8k out · 120s |
GPT-6 Sol gpt-6-sol | OpenAI | $2.00 / $10.00 | 100k in · 8k out · 120s |
GPT-6.1 Sol gpt-6.1-sol gpt-sol | OpenAI | $2.00 / $10.00 | 100k in · 8k out · 120s |
GPT-5.6 Terra gpt-5.6-terra gpt-terra | OpenAI | $2.00 / $12.00 | 100k in · 8k out · 120s |
GPT-5.6 Sol gpt-5.6-sol | OpenAI | $4.00 / $20.00 | 100k in · 8k out · 180s |
GPT-6 Astra gpt-6-astra gpt-astra | OpenAI | $10.00 / $50.00 | 100k in · 8k out · 180s |
Text Image 8 models
One 1024×1024 image per run from a text prompt, billed per image.
| Model | Provider | per image · list | Caps |
|---|---|---|---|
FLUX.2 Klein 4B (Workers AI) @cf/black-forest-labs/flux-2-klein-4b | Workers AI | ~$0.0011 | 1 image · 60s |
FLUX.1 Schnell (Workers AI) @cf/black-forest-labs/flux-1-schnell | Workers AI | ~$0.0019 | 1 image · 60s |
FLUX.2 Klein 9B (Workers AI) @cf/black-forest-labs/flux-2-klein-9b flux-klein | Workers AI | ~$0.015 | 1 image · 60s |
Leonardo Phoenix 1.0 (Workers AI) @cf/leonardo/phoenix-1.0 | Workers AI | ~$0.034 | 1 image · 60s |
Leonardo Lucid Origin (Workers AI) @cf/leonardo/lucid-origin | Workers AI | ~$0.036 | 1 image · 60s |
GPT Image 2 gpt-image-2 | OpenAI | ~$0.060 | 1 image · 180s |
GPT Image 2.5 Flare gpt-image-2.5-flare gpt-image | OpenAI | ~$0.060 | 1 image · 180s |
GPT Image 2.5 Sunburst gpt-image-2.5-sunburst | OpenAI | ~$0.060 | 1 image · 180s |
Text Speech 4 models
Text-to-speech: the run's text field (up to 2,000 characters) comes back as one MP3, billed per character.
| Model | Provider | per 1k characters · list | Caps |
|---|---|---|---|
MeloTTS (Workers AI) @cf/myshell-ai/melotts | Workers AI | $0.000205 | 1 audio file · 60s |
Deepgram Aura 1 (Workers AI) @cf/deepgram/aura-1 | Workers AI | $0.015 | 2,000 chars · 120s |
Deepgram Aura 2 English (Workers AI) @cf/deepgram/aura-2-en aura | Workers AI | $0.030 | 2,000 chars · 120s |
Deepgram Aura 2 Spanish (Workers AI) @cf/deepgram/aura-2-es | Workers AI | $0.030 | 2,000 chars · 120s |
Audio Text 1 model
Speech-to-text: one audio file attached with $files comes back as a transcript, billed per minute of audio.
| Model | Provider | per audio minute · list | Caps |
|---|---|---|---|
Whisper Large v3 Turbo (Workers AI) @cf/openai/whisper-large-v3-turbo whisper | Workers AI | $0.000513 | 1 audio file · 180s |
Pin a concrete id and you own the upgrades; a tier alias moves with the catalogue.
"Text + image" is claimed only for models that named the colour of a test image
through the same request shape a run uses — a text-only model handed an image fails
the run and refunds it. Models this deployment cannot run are not shown;
GET /v1/models still catalogues them with available: false
and carries the same input_types / output_type fields.
How server pricing works
- Pay for what a run actually uses. Each run is charged on the tokens the model actually consumed, rounded up to whole credits (minimum 1 credit ≈ $0.0001). There is no fixed per-job overhead, so a short turn on a cheap model bills a handful of credits.
- The model is pass-through; the fee pays for the servers. A run's base is the provider's list cost — charged exactly as the provider publishes it, no markup and no overhead. SkillSafe adds a 10% platform fee on that base for the computation and API handling it does on its own servers; the creator takes their markup — up to 100% — of the very same base, shown on each app's detail page. At the default 10% the two are equal, and nothing is deducted from the creator's side: they keep 100% of their markup.
- Prompt caching is passed through too. OpenAI caches any prompt of
1,024 tokens or more on its own and reports how many input tokens it served from
the cache. Those are billed at OpenAI's cached-input price, 10% of the input price.
On GPT-5.6 and GPT-6, the first run that stores a prompt is billed at the cache-write
price (125%) for the stored part. An app whose system prompt stays the same across
runs therefore pays much less for input after the first run.
GET /v1/modelslists both prices per model. - BYOK apps bill the 1-credit minimum ($0.0001) per run. When a publisher brings their own provider key, their key pays the provider directly and the platform charges only the minimum billable unit — no 10% platform fee and no creator markup.
- Spend is capped per job. Every model carries hard caps on input tokens, output tokens, and wall-clock time, so a single run can never overrun its hold.
- Image models bill per image. An image-generation run produces one 1024×1024 image. The hold reserves the worst-case cost of that image and the run settles down to the provider's actual (usage-reported where available) cost — same base, same two cuts as text runs.
- No model configured? Apps without a model run on the Workers AI default — cheap, keyless, and never silently billed at premium rates.